Supplementary Information for In silico QSAR and design of chalcone derivatives for HT-29 colorectal cancer: MLR and ANN approaches

This repository contains the supplementary data, raw datasets, and computational workflows supporting the research article: "In silico QSAR and design of chalcone derivatives for HT-29 colorectal cancer: MLR and ANN approaches" by Tony Nyo et al., published in Discover Chemistry. The study presents a comparative analysis of Multiple Linear Regression (MLR) and Artificial Neural Network (ANN) models to predict the anticancer activity of 193 chalcone derivatives against the HT-29 colorectal cancer cell line. The dataset is provided to ensure transparency, reproducibility, and validation of the QSAR models described in the main publication. Contents of this Repository: 1. Supplementary Tables:- Table S1: Complete dataset of 193 chalcone derivatives with SMILES notations and experimental pIC50 values.- Table S3: ANOVA statistics for the Stepwise MLR model.- Table S4: Correlation matrices showing inter-correlation between the 27 selected molecular descriptors.- Table S5: Comparative performance data for all tested ANN architectures (13i–4N–1O to 13i–10N–1O).- Table S6: Standardized regression coefficients and statistical significance for all 27 descriptors.- Table S7: Detailed Wilcoxon Signed-Rank Test results comparing absolute errors of MLR and ANN models. 2. Supplementary Files:- File S1: Full library of designed compounds with their calculated descriptors and predicted activities.- File S2: Full statistical calculation workflows for all validation parameters (R², Q², RMSEP, MAPE, etc.). 3. Supplementary Figures:- Figure S1: ANN Architecture (13i–8N–1O) illustrating the optimized feed-forward multilayer perceptron configuration.- Figure S2: Full ProTox-III Toxicity Prediction Profiles for the lead candidate Modifikasi_W_136. Funding:This research received no specific external grant funding. Software licenses and computational resources were provided through institutional support at Lambung Mangkurat University, Indonesia. Related Publication:Nyo, T., Triyasmono, L., & Santoso, U. T. (2025). In silico QSAR and design of chalcone derivatives for HT-29 colorectal cancer: MLR and ANN approaches. Discover Chemistry. License:Creative Commons Attribution 4.0 International (CC BY 4.0).本仓库包含支撑研究论文《In silico QSAR and design of chalcone derivatives for HT-29 colorectal cancer: MLR and ANN approaches》（作者Tony Nyo等，发表于《Discover Chemistry》）的补充数据、原始数据集与计算工作流。 本研究针对193种查尔酮衍生物开展对比分析，构建多元线性回归（Multiple Linear Regression, MLR）与人工神经网络（Artificial Neural Network, ANN）模型，以预测其对HT-29结直肠癌细胞系的抗癌活性。本数据集的发布旨在保障主刊中所述定量构效关系（Quantitative Structure-Activity Relationship, QSAR）模型的透明度、可复现性与验证性。 本仓库内容如下： 1. 补充表格： - 表S1：193种查尔酮衍生物的完整数据集，包含简化分子线性输入规范（Simplified Molecular Input Line Entry System, SMILES）标注与实验pIC₅₀值。 - 表S3：逐步多元线性回归模型的方差分析（Analysis of Variance, ANOVA）统计结果。 - 表S4：相关性矩阵，展示27个精选分子描述符间的相互关联。 - 表S5：所有测试人工神经网络架构（13i–4N–1O 至 13i–10N–1O）的对比性能数据。 - 表S6：27个分子描述符的标准化回归系数与统计学显著性结果。 - 表S7：对比多元线性回归与人工神经网络模型绝对误差的详细Wilcoxon符号秩检验结果。 2. 补充文件： - 文件S1：设计化合物的完整库，包含其计算得到的分子描述符与预测活性数据。 - 文件S2：所有验证参数（决定系数R²、交叉验证相关系数Q²、均方根误差RMSEP、平均绝对百分比误差MAPE等）的完整统计计算工作流。 3. 补充图片： - 图S1：人工神经网络架构（13i–8N–1O），展示优化后的前馈多层感知器配置。 - 图S2：候选先导化合物Modifikasi_W_136的完整ProTox-III毒性预测概况。 资助说明：本研究未获得特定外部项目资助，软件许可与计算资源由印度尼西亚兰邦曼库拉特大学提供机构配套支持。 相关已发表文献：Nyo, T., Triyasmono, L., & Santoso, U. T. (2025). In silico QSAR and design of chalcone derivatives for HT-29 colorectal cancer: MLR and ANN approaches. Discover Chemistry. 许可协议：知识共享署名4.0国际许可（Creative Commons Attribution 4.0 International, CC BY 4.0）。